11-11-2026 
|
Getting your Trinity Audio player ready...
|
TL;DR: The web was built around humans clicking links, visiting pages, viewing ads, and buying things. AI is already weakening that model by answering questions without sending users to the original source. Agentic commerce takes the next step by allowing machines to discover products, call services, and pay directly. If machines become major users of the web, publishers and service providers will need to monetize machine access through APIs, HTTP 402, and micropayments.
Key Takeaways:
- AI overviews and other answer engines are reducing the number of clicks through to source websites, weakening the advertising, subscription, and sales revenue that funds much of the web.
- AI agents will interact with websites differently from humans, often preferring structured data, APIs, and machine-readable actions over pages and buttons.
- HTTP 402 Payment Required provides one possible model for charging machines directly for digital resources.
- If AI agents make huge numbers of tiny payments, the payment infrastructure which underpins it will need extremely high throughput with reliably low costs.
- The web doesn’t need to force AI agents to act like humans; it needs an economic model designed for machines at scale.
The original web was built around clicks.
The bargain so far has been simple—publishers create useful content, Google (NASDAQ: GOOGL) and other search engines index it, and if someone searches for an answer, they click a result and visit a website or watch a video. The publisher monetizes through advertising, subscriptions, sales, affiliate links, and other methods.
However, generative artificial intelligence (AI) is beginning to weaken that agreement. AI overviews and other answer engines can extract information from several websites, combine it into one response, and answer the user without requiring them to visit original sources. One study found that users click links inside AI overviews only 1% of the time.
For users, this can be incredibly convenient, but for the websites creating the information, it creates an obvious economic problem.
“The reality is the AI Mode does not drive traffic, not for us and not to anybody as far as I know. So it’s just a fundamental shift in how Google works. It used to be a traffic driver. Now it’s not.” – Reddit CEO Steve Huffman on the Q3 2026 Earnings Call
The web’s old bargain is beginning to break
AI systems depend on fresh information, but the sites they depend on for it also have their own requirements. News sites need reporters, product databases need updating, technical documentation has to change as software changes, and researchers need to publish new findings. Specialists need a financial incentive to spend time researching and publishing accurate information.
Traditionally, traffic helped pay for all of this, but as stated above, traffic is already collapsing in the face of AI overviews and other AI sources that send fewer people back to the source.
The incentive problem is obvious—the better AI systems become at answering questions without sending users elsewhere, the less reason many publishers have to create that information in the first place.
Mike Hearn raised this point in a CoinGeek interview with Kurt Wuckert Jr. If AI replaces traditional search traffic, websites may lose the advertising and other revenue that traffic helped generate.
One potential solution is for the machine consuming the information to pay for it directly, but this would require thinking about the web very differently.
AI agents don’t browse like humans
Fundamentally, most websites are designed around human behavior. This involves users landing on homepages, navigating menus, reading product pages, filling out forms, entering payment information, and clicking buttons.
AI agents don’t necessarily do or require any of this. An agent trying to book a hotel, buy inventory, access research, or query a database may simply need to discover what is available, authenticate itself, request the resource, pay for it, and get the desired result.
Cloudflare (NASDAQ: NET) describes this emerging model as the Agentic Internet, in which websites and services become discoverable, readable, callable, and payable by machines as well as people.
The difference is obvious—a human looking for a flight might search Google, visit a travel website, choose a hotel, and make a booking. Along the way, they might have seen a banner ad which helps the website generate revenue.
An AI agent can extract all of the necessary information, compare offers, and make bookings without a human ever visiting a website directly. In this way, the web page itself becomes less important, and the underlying product or service becomes the product.
This could apply to almost anything online, from articles and API calls to database queries, tool calls, map requests, and product listings.
And if machines increasingly access resources directly, creators will need to charge machines directly, too.
HTTP 402 could finally have its moment
The web already has a status code for this. HTTP 402 Payment Required has existed for decades but was reserved for future use.
The basic concept is simple: a machine requests a resource, the server responds that payment is required, the machine pays, and the server provides access.
This removes several steps designed for humans, such as creating an account, selecting a subscription plan, entering card details, and moving through a checkout page.
Cloudflare is already experimenting with this model through its Monetization Gateway. It uses HTTP 402 to let websites charge AI agents for access to content, tools, APIs, and datasets.
This changes the economics of access more generally. Instead of asking users to subscribe to a $10 per month plan, an AI agent may pay $0.01 for access to a single article. Likewise, an API provider might charge a fraction of a cent per call rather than monetizing via ads and sponsorships.
In this model, machines would pay for exactly what they consume.
Micropayments could restore the incentive to publish
The micropayments model is particularly interesting for creators and publishers.
Imagine an AI agent needs information from five different sources to answer a question. Under the current model, it may crawl five websites, extract the relevant information, synthesize an answer, and provide citations. However, as the data already shows, very few users actually visit those sources.
Under a micropayments model in which machines pay for access, the AI would make tiny payments to access information. The user would still get the answer immediately, the AI system would still get access to high-quality, accurate information, and the webmaster or creator would get paid for providing it.
This wouldn’t replace every existing web business model. Advertising, subscriptions, affiliate marketing, and direct sales would almost certainly continue to exist, but machine payments would create an additional revenue stream for creators, database administrators, webmasters, and others.
In this way, the incentive problem we identified earlier is partially solved. The people who create the content the AI systems depend on would have a financial incentive to continue producing it.
Machine customers need machine-scale payments
One obvious challenge to this model is that traditional payment systems weren’t designed for it and thus aren’t capable of facilitating payments at this scale.
Credit cards are great when a human being is making a $50 purchase, and PayPal (NASDAQ: PYPL) is fine as long as the fees aren’t greater than the payment itself. However, neither is suitable for a machine economy with millions of transactions per second (TPS) worth fractions of a cent.
This is where digital currencies on scalable public blockchains can provide a solution. AI agents could hold a wallet, receive payment instructions, send small amounts to a service provider, and gain access to a resource without a traditional payment intermediary.
At machine scale, the requirements become obvious—transaction fees need to remain extremely low, payments need to settle quickly, and the network needs to be able to process huge volumes of payments without becoming a bottleneck. A machine economy making billions of tiny payments requires infrastructure up to the task.
The web doesn’t need to force AI agents to act like humans; it needs an economic model designed for machines at scale.
FAQs:
What is HTTP 402 Payment Required?
HTTP 402 Payment Required is an old HTTP status code reserved for situations where access to a digital resource requires payment. New machine payment systems are beginning to use it to enable software and AI agents to pay directly for websites, APIs, datasets, and other resources.
How are AI overviews affecting web traffic?
AI overviews can answer questions directly on search results pages, reducing the need to visit the websites providing the underlying information. Several studies have found substantially lower click-through rates when AI-generated answers appear.
Why would AI agents need micropayments?
AI agents can consume small amounts of information or services from many different sources or providers. Micropayments allow them to pay only for the specific resources they use without requiring subscriptions or conventional payments.
How could blockchain support machine payments?
A scalable public blockchain can allow AI agents to make direct digital payments, including very small transactions, while maintaining a shared record of settlement. This can support machine-to-machine payments without relying on traditional payment flows for every interaction.
Will AI agents replace websites?
Not necessarily. Websites will continue serving humans, but their underlying services may increasingly expose structured data, APIs, machine-readable actions, and payment endpoints for AI agents as well as human users.
In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.
Watch: Can we trust AI? How blockchain and IPv6 could fix accountability
